Cav 2001
Cav 2001
002 1
Abstract
A pressure-based algorithm is developed and applied to compute turbulent sheet cavitating flows. Single-fluid
Navier-Stokes equations, cast in their conservative form, along with a volume fraction transport equation are
employed. The flow is computed in both phases with the vapor pressure recovered inside the cavity via a mass
transfer model. A pressure-velocity-density coupling scheme along with an upwinded density interpolation is
developed to handle the large density ratio associated with cavitation. The method is assessed through simulations of
cavitating flows over a cylindrical object and an airfoil. The results show satisfactory agreement with experimental
data in pressure distribution. In addition, information such as wall shear stress distributions and related velocity and
turbulence fields is highlighted for both axisymmetric projectile and NACA airfoil.
1 Introduction
Computational modeling of cavitation has been pursued for years. Early studies primarily utilize the potential flow
theory; they are still widely used in many engineering applications. Studies dealing with cavitation modeling
through the computation of the Navier-Stokes (N-S) equations have emerged in the last decade. A table
summarizing some of the selected studies is included in the appendix. A review of these studies is presented in
Senocak and Shyy (2001). To account for the cavitation dynamics in a more flexible manner, recently, a transport
model is developed. In this approach volume or mass fraction of liquid (and vapor) phase is convected. Singhal et al.
(1997), Merkle et al. (1998) and Kunz et al. (1999, 2000) have employed similar models based on this concept with
differences in the source terms. One apparent advantage of this model comes from the convective character of the
equation, which allows modeling of the impact of inertial forces on cavities like elongation, detachment and drift of
bubbles. Merkle et al. (1998) and Kunz et al. (1999, 2000) have employed the artificial compressibility method.
Kunz et al (1999, 2000) have adopted a non-conservative form of the continuity equation and applied the model to
different geometries. Their solutions are in good agreement with experimental measurements of pressure
distributions. In these studies special attention has been given to the preconditioning formulation in order to create a
robust artificial compressibility method
So far, in the open literature, there seems to be a lack of pressure-based methods for computing cavitating
flows. By pressure-based method, we mean that the pressure field is solved by combining the momentum and mass
continuity equations to form a pressure or pressure-correction equation (Patankar, 1980; Shyy, 1994) In the present
study, a pressure-based algorithm with conservative formulation, multi-block, curvilinear grid systems, is adopted to
compute cavitating flows. In particular, the coupling between velocity, pressure and density, for proper formulation
of the pressure correction equation for cavitating flow conditions will be discussed. The mass transport equation
cavitation model, such as that employed by Kunz et al. (1999, 2000) will be adopted.
In what follows, we first present the governing equations and main features of the cavitation model, and then
propose numerical schemes that ensure stable numerical computations. The presented results include simulations of
both noncavitating and cavitating flows around a cylindrical object with hemispherical headform and a NACA0012
airfoil.
CAV2001:sessionA7.002 2
2 Theoretical Formulation
The set of governing equations consists of the conservative form of the Reynolds averaged Navier-Stokes equations,
plus a volume fraction transport equation to account for the cavitation dynamics. The equations, written in the
Cartesian coordinates for the ease of presentation, are presented below.
∂ρ m ∂( ρ m u j )
+ =0 (1)
∂t ∂x j
∂ ∂ ∂p ∂ ∂u ∂u
( ρ m ui ) + ( ρ m ui u j ) = − + ( µ + µ t ) i + j (2)
∂t ∂x ∂x ∂x ∂x ∂x
j i j j i
∂α l ∂ + −
+ (α u ) = (m& + m& ) (3)
∂t ∂x j l j
The mixture density and the turbulent viscosity are defined, respectively, as follows:
2
ρ mCµ k
ρ m = ρ l α l + ρ v (1 − α l ) µt = (4)
ε
For the turbulence closure, the original k-ε turbulence model with wall functions is adopted (Jones and Launder,
1972).
Cavitation modeling
Physically, the cavitation process is governed by the thermodynamics and the kinetics of the phase change dynamics
− +
occurring in the system. This complex phenomenon is modeled through m& and m& terms in Eq. (3), which
represent evaporation and condensation of the phases, respectively, and results in a variable density field. Surface
tension and buoyancy effects are neglected considering the typical situation that Weber and Froude numbers are
large. The particular form of these phase transformation rates are adopted from Kunz et al. (1999). The values of the
empirical constants Cdest and Cprod for each simulation are presented along with corresponding figures and they are
different than the values reported in other studies using the same cavitation model. The sensitivity of the simulations
to these constants is also studied. The source terms that are adopted in this study are given below:
C prod ρ vα l (1 − α l )
2
C ρ α MIN [0, p − p v ]
− +
m& = dest v l m& = (5)
1 2 ρ l t∞
ρ l ρ lU ∞ t ∞
2
The time scale in the equation is defined as the ratio of the characteristic length scale to the reference velocity
scale (l/U). The nominal density ratio (ρl/ρv) is the ratio between thermodynamic values of density of liquid and
vapor phases at the corresponding flow condition; a value of 1000 is taken for this ratio in all computations in this
study.
3 Numerical Method
The present Navier-Stokes solver, documented in Shyy (1994), Shyy et al. (1997) and Thakur et al. (1997) employs
a pressure-based algorithm and a finite volume approach to solve the fluid flow and energy equations, on multi-
block structured curvilinear grids in 2D and 3D domains. For the present cavitation model, Eq.(3), the volume
fraction transport equation with appropriate source terms given in Eq.(5), needs to be implemented into the solver.
CAV2001:sessionA7.002 3
For further details of the pressured based method for cavitating flows the reader is referred to Senocak and Shyy
(2001).
Pressure-velocity-density coupling
In the pressure-based algorithm, the pressure correction equation has been revised to achieve successful solutions
for highly compressible flows (Shyy and Braaten 1988; Karki and Patankar 1989). In the cavitation model a
convection equation with pressure dependent source terms, Eq. (3), is solved to determine the density field. Because
of this coupling between pressure and density, the pressure correction equation needs to be reformulated, even
though the Mach number effect is not explicitly addressed in the model. Once the cavitation model is implemented
into a pressure-based algorithm, the pressure correction equation exhibits a convective-diffusive nature in cavitating
regions and purely diffusive nature in the liquid phase. In the present algorithm, the following relation between
density and pressure is introduced to establish the pressure-velocity-density coupling.
ρ ′ = C (1 − α l ) P ′ (6)
where C is an arbitrary constant. It should be emphasized that the choice of this constant does not affect the final
converged solution because of the nature of the pressure correction equation. It is found that a very large value for C
can destabilize the computation in early stage of the iteration process. For this reason, we suggest C=O(1) be used.
In our computations, C=4 is adopted. The above scheme results in a combined incompressible-compressible
formulation that preserves the incompressible nature in the liquid phase. In the cavitating region, it accounts for the
pressure-density dependency in a nonlinear fashion, in accordance with the local value of αl. This modification is
key to a stable computation in which the uniform vapor pressure is recovered in the final converged solution.
Another aspect is that, similar to compressible flow computations, the density at the cell face is upwinded
(Shyy, 1994). The criterion for upwinding is based on the value of liquid volume fraction; that is, wherever αl is less
than 1.0, the cell-faced density value is estimated based on an upwinded formula. This treatment significantly
improves the convergence level and has a stabilizing effect in the vicinity of sharp density gradients.
It should also be emphasized that Eq. (6) is not limited to the cavitation model employed in this study; it can
easily be adopted for other cavitation models. For example, if an equation of state is utilized to generate the variable
density field, then vapor or mass fraction can be derived from density values and used in Eq. (6) to establish the
pressure-density coupling.
are effective on more grid points. The cavity detachment point remains fixed in both of the simulations, which is
also in agreement with experimental data.
In Figure 2, the sensitivity of the solutions to model parameters is studied. It can be seen that even increasing
these parameter by an order of magnitude has little effect on the pressure coefficient predictions. However, the
computed density ratio is noticeably different between these model parameters. Clearly, the computed density ratios
can be controlled through adjustment of the model parameters to yield very different solutions while pressure
predictions remain little unaffected. The density profiles indicate a sharp discontinuity at the closure region with a
reentrant jet located downstream of it, which possibly explains the reason of localized surface erosion in the closure
region of sheet cavities.
In Figure 3 the effect of cavitation on wall shear stress distribution is studied. By comparing the skin friction
coefficient of both noncavitating and cavitating conditions, one can see that the existence of cavitation not only
alters the flow structure inside the cavity but it also affects the downstream flow. The turbulent viscosity
distributions indicate that the reentrant jet gets more dissipative as the cavitation number is lowered. This suggests
that the viscous effects can play an important role on the overall cavity behaviour such as the reattachment location.
Hemispherical, Re=1.36x105
1.2
noncavitating
1 σ=0.40
σ=0.30
Exp. noncavitating
0.8
Exp. σ=0.40
Exp. σ=0.30
σ=0.40
0.6
ρ l ρ min =47
0.4
Cp
0.2
-0.2
σ=0.30
-0.4
ρ l ρ min =653
-0.6
-0.8
0 1 2 3 4 5
s/d
Figure 1. Comparison of pressure coefficient distributions for hemispherical object under noncavitating and
cavitating conditions (Cdest=9x105, Cprod=3x104, ρl/ρv=1000). Experimental data is from Rouse and McNown (1948).
Hemispherical, σ=0.40, Re=1.36x105
1.2
1
density
Cdest=2x106, Cprod=1.3x105
0.8
Cdest=9x105, Cprod=3x104 Cdest=9x105, Cprod=3x104
0.6 Exp. data ρ ρ min =47
Cp, density
l
0.4
0.2
-0.2
Cdest=2x106, Cprod=1.3x105
ρ ρ min =739
-0.4 l
-0.6
0 0.5 1 1.5 2 2.5 3 3.5 4
s/d
Figure 2. Sensitivity of modeling parameters for the hemispherical object at σ=0.40 (ρl/ρv=1000). Experimental
data is from Rouse and McNown (1948).
pressure distribution along with the cavity shape is plotted in Figure 3. The angle of attack is 1° and the cavity
occurs at mid chord with a corresponding cavitation number of 0.42. These conditions and the overall behavior of
the cavity are consistent with the experimental study of Shen and Dimotakis (1989) in which the NACA66MOD
airfoil is investigated. The corresponding vapor pressure is successfully recovered inside the cavity region that is
also consistent with our results of hemispherical object. Unlike the hemispherical object no reentrant jet is observed
in this case, possibly because the cavity is in the form of a thin layer.
0.012
noncavitating
σ=0.30
σ=0.40
0.01
Noncavitating
0.008
0.006
Cf
0.004
σ = 0.40
0.002
0
0 0.5 1 1.5 2 2.5 3 3.5 4 σ = 0.30
s/d
Figure 3. Effect of cavitation on wall shear stress. Corresponding turbulent viscosity distributions are on the right
part of the figure.
Suction side
Figure 4. Sheet cavitation on NACA 0012 airfoil at σ=0.42, ρl/ρmin=8 (Cdest=9x105, Cprod=3x104, ρl/ρv=1000).
CAV2001:sessionA7.002 6
5 Conclusions
Single-fluid Navier-Stokes equations, cast in their conservative form, along with a volume fraction transport
equation are employed to model cavitating flows over a cylindrical object and a NACA0012 airfoil. The flow is
computed in both phases with the vapor pressure recovered inside the cavity via a mass transfer model. A pressure-
velocity-density coupling scheme is developed and implemented into a pressure-based algorithm to compute
cavitating flows. The proposed coupling scheme along with density upwinding for cavitating regions is the key to
stable computations of cavitating flows.
Combined with the multiblock and curvilinear grid systems, the present flow solver can handle large density
ratios and complex geometries. For the turbulent flows with sheet cavitation, the density profiles indicate a sharp
discontinuity at the closure region with a reentrant jet located downstream of it. As the cavitation number is lowered
the reentrant jet gets stronger and more dissipative. While the pressure distribution is less sensitive to model
parameters, density distribution exhibits a higher sensitivity to them. Identical cavitation model parameters are used
for the hemispherical object and the NACA0012 airfoil suggesting that the present cavitation can be employed for
further applications.
The future work will concentrate on applying the method to simulate different forms of cavitation such as cloud
and supercavitation, and to investigate the flow physics to gain better understanding.
Acknowledgements
This study has been supported partially by ONR and NSF. The authors also would like to thank Robert F. Kunz and
Jules.W. Lindau for helpful discussions on the cavitation model.
NOMENCLATURE
Symbols
C arbitrary constant
Cdest, Cprod empirical constants in Eq.(5)
CP pressure coefficient
−
m& evaporation rate
+
m& condensation rate
ui velocity in Cartesian coordinates
xi Cartesian coordinates
P pressure
P′ pressure correction
Uo, U∞ u-velocity at a reference point
t, t∞ time, mean flow time scale
α volume fraction
µ laminar viscosity
µt turbulent viscosity
νt kinematic viscosity
ρm mixture density
ρ′ density correction
σ cavitation parameter
Subscripts, Superscripts
l liquid phase
v vapor phase
∞ freestream
References
Ahuja, V., Cavallo, P.A. and Hosangadi, A. (2000) “Multi-Phase Flow Modeling on Adaptive Unstructured
Meshes,” AIAA Fluids 2000 and Exhibit, Paper No. AIAA-2000-2662, June 19-22, Denver, Colorado.
Brennen, C.E. (1995), Cavitation and Bubble Dynamics, Oxford University Press, New York.
CAV2001:sessionA7.002 7
Chen, Y. and Heister, S.D. (1994) “A Numerical Treatment for Attached Cavitation,” Journal of Fluids
Engineering, Vol. 116, pp. 613-618.
Chen, Y. and Heister, S.D. (1996) “Modeling Hydrodynamic Nonequilibrium in Cavitating Flows,” Journal of
Fluids Engineering, Vol. 118, pp. 172-178.
Deshpande, M., Feng, J. and Merkle, C.L. (1997) “Numerical Modeling of the Thermodynamic Effects of
Cavitation,” Journal of Fluids Engineering, Vol. 119, pp. 420-427.
Edwards, J.R., Franklin, R.K. and Liou, M.S. (2000) “Low-Diffusion Flux-Splitting Methods for Real Fluid
Flows with Phase Transitions,” AIAA Journal, Vol. 38, No. 9, pp. 1624-1633.
Gopalan, S. and Katz, J. (2000) “Flow Structure and Modeling Issues in the Closure Region of Attached
Cavitation,” Phys. Fluids, Vol. 12, No. 4, pp. 895-911.
Jones, W.P. and Launder, B. E. (1972) “The Prediction of Laminarization with a Two-Equation Model of
Turbulence,” Int. J. Heat Mass Trans., Vol. 15, pp. 301-314.
Karki, K.C. and Patankar S.V. (1989) “Pressure Based Calculation Procedure for Viscous Flows at All Speeds
in Arbitrary Configurations,” AIAA Journal, Vol. 27, No. 9, pp. 1167-1174.
Knapp, R. T., Daily, J.W. and Hammitt, F. G. (1970), Cavitation, McGraw-Hill, New York.
Kubota, A., Kato, H. and Yamaguchi, H. (1992) “A New Modelling of Cavitating Flows: A Numerical Study of
Unsteady Cavitation on a Hydrofoil Section,” J. Fluid Mech., Vol. 240, pp. 59-96.
Kunz, R.F., Chyczewski, T.S., Boger, D. A., Stinebring, D. R. and Gibeling, H. J. (1999) “Multi-Phase CFD
Analysis of Natural and Ventilated Cavitation About Submerged Bodies,” ASME Paper FEDSM99-7364,
Proceedings of 3rd ASME/JSME Joints Fluids Engineering Conference.
Kunz, R.F., Boger, D.A., Stinebring, D.R., Chyczewski, T.S., Lindau, J.W., Gibeling H.J., Venkateswaran, S.
and Govindan, T.R. (2000) “A Preconditioned Navier-Stokes Method for Two-phase Flows with Application to
Cavitation Prediction,” Computers & Fluids, Vol. 29, pp. 849-875.
Merkle, C.L., Feng, J. and Buelow, P.E.O. (1998) “Computational Modeling of the Dynamics of Sheet
Cavitation,” 3rd International Symposium on Cavitation, Grenoble, France.
Patankar, S.V. (1980), Numerical Heat Transfer and Fluid Flow, Hemisphere, Washington DC.
Senocak, I. and Shyy, W. (2001) “A Pressure-Based Method for Turbulent Cavitating Flow Computations,” 31st
AIAA Fluid Dynamics Conference and Exhibit, AIAA 2001-2907. Also available on the web
http://www.aero.ufl.edu/cfd/
Shen, Y. and Dimotakis, P. (1989) “The Influence of Surface Cavitation on Hydrodynamic Forces,” Proc. 22nd
ATTC, St. Johns, pp. 44-53.
Shyy, W. and Braaten, M.E. (1988) “Adaptive Grid Computation for Inviscid Compressible Flows Using a
Pressure Correction Method, Proceedings of the AIAA/ASME/SIAM/APS First National Fluid Dynamics Congress,
AIAA-CP 888, pp. 112-120.
Shyy, W. (1994), Computational Modeling for Fluid Flow and Interfacial Transport, Elsevier, Amsterdam, The
Netherlands, (revised printing 1997).
Shyy, W., Thakur, S.S., Ouyang, H., Liu, J. and Blosch, E. (1997), Computational Techniques for Complex
Transport Phenomena, Cambridge University Press, New York.
Singhal, A.K., Vaidya, N. and Leonard, A.D. (1997) “Multi-dimensional Simulation of Cavitating Flows Using
a PDF Model for Phase Change,” ASME Paper FEDSM97-3272, The 1997 ASME Fluids Engineering Division
Summer Meeting.
Rouse, H. and McNown, J.S. (1948) “Cavitation and Pressure Distribution, Head Forms at Zero Angle of Yaw,”
Studies in Engineering, Bulletin 32, State University of Iowa.
Thakur, S.S., Wright, J.F., Shyy, W. and Udaykumar, H. (1997) “SEAL: A Computational Fluid Dynamics and
Heat Transfer Code for Complex 3-D Geometries,” University of Florida, Gainesville.
Venkateswaran, S., Lindau, J.W., Kunz, R.F. and Merkle, C.L. (2001) “Preconditioning Algorithms for the
Computation of Multi-Phase Mixture Flows,” AIAA 39th Aerospace Sciences Meeting & Exhibit, Paper No. 2001-
0125.
Ventikos, Y., and Tzabiras, G. (2000) “A Numerical Method for the Simulation of Steady and Unsteady
Cavitating Flows,” Computers & Fluids, Vol. 29, pp. 63-88.
CAV2001:sessionA7.002 8
Appendix
Table 1. Overview of selected studies on numerical simulations of cavitating flows based on the solution of N-S
equations.
NUMERICAL APPLICATIONS/MAIN
AUTHORS CAVITATION MODEL
ALGORITHM FINDINGS
Rayleigh-Plesset (R-P) equation is
coupled to the Poisson equation. Marker and Cell Cloud cavitation on hydrofoils.
Kubota et al. (1992 Cavity region is modeled as 3-D N-S equations Numerical instability for high-
compressible fluid with variable No turbulence model. density ratio. Re=3x105
density
Marker and Cell Pressure distribution on
Chen and Heister Interface tracking based on P=Pvap
2-D N-S equations axisymmetric geometries.
(1994) Grid conforms to the cavity shape.
No turbulence model Re=1.36x105
Marker and Cell Pressure distribution on
Chen and Heister Time and pressure dependent
2-D N-S equations axisymmetric geometries.
(1996) pseudo-density equation.
No turbulence model Re=1.36x105
Interface tracking based on P=Pvap Artificial Compressibility Sheet cavitation for cryogenic
Deshpande et al.
with mass transfer. 2-D N-S equations fluids. Studied the thermal
(1997)
Grid conforms to the cavity shape No turbulence model boundary layer over cavity.
Artificial Compressibility
Merkle et al. Vapor mass fraction equation with 2-D N-S equations Pressure distribution on
(1998) pressure dependent source terms. Two equation turbulence hydrofoils.
model
Volume fraction equation with
Artificial Compressibility Pressure distribution on
Kunz et al. (1999, pressure dependent source terms
3-D N-S equations axisymmetric geometries.
2000) Nonconservative continuity
k-ε turbulence model Re=1.36x105
equation. Preconditioning strategy.